Human liver cancer, with increasing occurrence in the United States, is the 5th most prevalent malignant disease in the world. It is the fourth leading cause of cancer mortality, which accounts for an estimated 1 million deaths annually. Hepatocellular carcinoma (HCC) is a major type of primary liver cancer. HCC is considered to be a terminally ill disease and currently, there is little progress toward the discovery of efficient therapies leading to regression. This is due largely to the lack of a method for early diagnosis and the lack of information on the phenotypic changes associated with the development of HCC. Our goals are to identify common gene clusters that are responsible for the genesis of HCC and to discover new genes critical for viral hepatitis-mediated HCC as well as genes necessary for metastasis. These studies will contribute to the establishment of novel markers with potential diagnostic and prognostic value, and analysis of these genes would provide further understanding of the genesis of liver cancer and provide further insights into designing strategies for HCC-directed molecular therapy. We have taken two approaches, namely, Serial Analysis of Gene Expression (SAGE) and cDNA microarray, to explore potential cellular genes that are expressed abnormally in primary human hepatocytes infected with the two viral hepatitis oncoproteins, HBx or HC-core, and in liver samples from chronic active hepatitis patients or HCC patients that differ in the status of HBV or HCV. In addition, we are comparing gene expression profiles between primary HCC and metastatic HCC. Infection of normal hepatocytes with HBx and HC-core is achieved by a replication-defective adenoviral vector. We have constructed two SAGE libraries derived from freshly isolated normal primary human hepatocytes with or without the expression of HBx(135). A total of 19,501 sequence tags (representing 1443 unique transcripts) were analyzed, which provide a distribution of a transcriptome characteristic of normal hepatocytes and a profile associated with HBx expression. Examples of the targeted genes were confirmed by the Megarray analysis with a significant correlation between quantitative SAGE and Megarray (r = 0.8, p<0.005). In HBx-expressing hepatocytes, a total of 57 transcripts (3.9%) were induced and 46 transcripts (3.3%) were repressed by more than 5-fold. There are nine novel upregulated genes (designated as XIG1-9) and 22 novel downregulated novel genes (designated as XSG1-22). Interestingly, among the known genes, most of the HBx-upregulated transcripts can be clustered into three major classes: genes that encode ribosomal proteins, transcription factors with zinc finger motifs, and proteins associated with protein degradation pathway. These results suggest that HBx may function as a major regulator in common cellular pathways that, in turn, regulate protein synthesis, gene transcription, and protein degradation. In addition, we have compared gene expression profiles in primary hepatocytes expressing HBx as well as liver samples from chronic liver diseases including HBV or HCV infection and in HCC. Clustering algorithms were used to identify the deregulation of distinctive gene expression profiles in these samples. Clustering algorithm analysis of the expression profiles indicates that there is a consistent alteration of a subset of oncogenes (such as c-myc and c-myb) and tumor suppressor genes (such as APC, p53, WAF1, and WT1). Many targeted genes were also found by the SAGE technique. Our findings are consistent with the hypothesis that the deregulation of cellular genes by oncogenic HBx may be an early event that favors hepatocyte proliferation during liver carcinogenesis. More recently, we analyzed the expression of 9,180 genes in a total of 67 HCC tumors from 40 patients without or with accompanying intra-hepatic metastases. Using a supervised machine learning algorithm to classify patients based on their gene expression profiles, we have generated for the first time a molecular signature that correctly classifies patients with or without metastases, and have identified genes that are mostly relevant to the prediction outcome including patient survival. Unexpectedly, we found that the gene expression signature of primary HCCs with accompanying metastasis was very similar to that of their corresponding metastases, implying that the genes favoring metastasis progression were likely initiated in the primary tumors. Moreover, osteopontin was overexpressed in primary HCC with intra-hepatic metastasis and a neutralizing antibody against osteopontin can block the invasion of highly metastatic HCC cells in an in vitro assay of invasion. Our studies offer a strategy to tailor HCC patients based on the gene expression profile to adjuvant therapy, and identify osteopontin both as a diagnostic marker and potential therapeutic target for metastatic HCC. We also are interested in the comparison of the molecular profiling of liver samples from many chronic liver diseases, including HBV, HCV, hemochromatosis, Wilson, alcohol, autoimmune chronic hepatitis, or primary biliary cirrhosis, with liver cancer. Our results demonstrate a unique gene expression pattern associated with these individual diseases. These results may offer insight into potential molecular mechanisms underlying precancerous conditions that are associated with HCC.